In general, a computer uses magnetic disk devices having high cost performance as devices for storing data. Because a magnetic disk is accompanied by mechanical operation, a processing speed is low (about 10 msec) as compared with that of a processor. Even if the processor is speeded up, unless the disk is speeded up, there are many cases where performance of the whole system is not improved. As a means for solving this problem, there is a disk array. As described in W. David Schwaderer, Andrew W. Wilson, Jr., “Understanding I/O Subsystems First Edition” pp. 271-291, the disk array is a technique for improving performance and reliability not only by arranging data on a plurality of drives in a distributed manner, but also by storing redundant data in the drives. In an especially large-scale system, performance and reliability are required at the same time. This is the reason why the disk array is used.
In recent years, a hopeful view is taken on a fiber channel as an interface for connecting between a disk array and a host. SCSI (Small Computer System Interface) can extend a connection distance only to several tens of meters, whereas the fiber channel can extend a connection distance to several kilometers. In addition, the number of devices, which can be connected to the fiber channel, is several times as many as the number of devices that can be connected to SCSI. As opposed to a local area network (LAN) used for data communication between hosts, a storage area network (SAN), which is used for data communication between a host and the other devices (for example, storage subsystems), receives attention in recent years. The fiber channel is also used as a communication medium of the storage area network. The fiber channel is standardized. If a device and a host comply with the standard, they can be connected to the storage area network. For example, it is possible to connect a disk array having a fiber channel interface to a plurality of hosts.
As a disclosed method for sharing data between a plurality of hosts, there is Japanese Patent Application Laid-Open No. Hei 9-325905 (hereinafter referred to as “document 1”). The document 1 discloses a disk device having a function of converting between a CKD (Count Key Data) format, which is a data format adopted in mainframes, and a FBA format that is a data format adopted in UNIX and the like. In this case, the format means structure of a minimum unit of data, which is transferred between a host and a disk device. The CKD is a variable-length format; and the FBA is a fixed-length format. In a disk, which stores data in the CKD format, user data, a data length, and a key used for searching are stored. For example, when writing user data having a length of 2048 bytes into a disk, a data length, a key value, and the data having a length of 2048 bytes are written at a time. On the other hand, as regards the FBA format, when writing user data having a length of 2048 bytes on a disk similarly, data having a length of 2048 Kbytes is stored in a disk on a basis of a divided block having a fixed length of 512 bytes. In the FBA, a data length and a key are not stored. As described above, even if the same disk drive is used, data storing formats used in the disk drive are different depending on kinds of hosts. Therefore, it is not possible to share data between different hosts. In order to solve this problem, the document 1 describes the following function: when a disk stored in the CKD format is accessed in the FBA format, converting a data portion of CKD into blocks in the FBA format.
As another file sharing technique, there is Japanese Patent Application Laid-Open No. Hei 9-179758 (hereinafter referred to as “document 2”). The following technology is disclosed in the document 2. Data are written into a data sharing area in a writable recording medium. In addition to it, a plurality of file tables, by which data management information, which is required when accessing data, can be referred to from two or more kinds of operating systems individually, are written into the recording medium.